Integrating Metabolomics and Proteomics to Reveal the Regulatory Network Governing the Natural Variation in Rice Seed Germination Rate

TL;DR

This study explores how metabolites like glutamine influence rice seed germination, revealing a key role in improving germination rates.

Contribution

The study identifies glutamine as a causal hub metabolite in rice seed germination through integrated metabolomic and proteomic analysis.

Findings

Amino acid, energy, and glutathione metabolism pathways are activated in rapid germination rice seeds.

Glutamine is significantly enriched in rapid germination seeds and promotes germination when applied externally.

Germination is controlled by a genotype-dependent regulatory network influenced by glutamine.

Abstract

Seed germination rate is a key early trait that strongly influences rice yield. Although germination is known to be regulated by classical phytohormones and certain metabolites, the systematic metabolic regulatory network underlying natural variation, especially the key hub metabolites with causal function, still lacks in-depth analysis. In this study, we investigated 56 rice accessions showing pronounced differences in germination performance and systematically identified metabolic pathways associated with germination rate by integrating metabolomic and proteomic analyses. Pathways involved in amino acid metabolism, energy metabolism, and glutathione metabolism were coordinately activated in Rapid Germination (RG) seeds compared with Delayed Germination (DG) seeds. Among them, glutamine was significantly enriched in the RG group. Exogenous application of glutamine selectively and significantly promoted radicle and shoot elongation in a subset of DG varieties, providing direct evidence for a positive causal role of glutamine in seed germination. The variety-specific response further suggests that germination is controlled by a complex, genotype-dependent regulatory network. Together, our results highlight a glutamine-centered metabolic program as an important basis for rapid rice seed germination and provide potential targets for improving early vigor through metabolic engineering and molecular breeding.

Figures

Detection and identification of specific LysoPC (16:1), L-glutamic acid-L-glutamine, and Quercetin-5-O-hexoside metabolite signals by Q-Exactive LC-MS/MS. (A) EIC (extracted ion chromatogram) of GR0715 at 8.6 min. (B) MS/MS spectra of GR0715 at m/z 494.3247, and the metabolite was identified as LysoPC (16:1). (C) The molecular structure of the LysoPC (16:1) and its general fragmentation rules. (D) EIC (extracted ion chromatogram) of GR0143 at 0.8 min. (E) MS/MS spectra of GR0143 at m/z 277.1036, and the metabolite was identified as L-glutamic acid-L-glutamine. (F) The molecular structure of the L-glutamic acid-L-glutamine and its general fragmentation rules. (G) EIC (extracted ion chromatogram) of GR0419 at 4.1 min. (H) MS/MS spectra of GR0419 at m/z 271.0981, and the metabolite was identified as Quercetin-5-O-hexoside. (I) The molecular structure of the Quercetin-5-O-hexoside and its general fragmentation rules. Std, standard; RG, Rapid Germination; DG, Delayed Germination.

Summary of metabolic profiling and tissue variability analysis. (A) Categorical pie chart of the 1386 annotated metabolites. (B) PCA results of the metabolome data from Rapid Germination (RG), Delayed Germination (DG) and quality control (QC) samples. (C) Hierarchically clustered heatmap of the 1386 annotated metabolites from 56 germinated seed samples.

Full text

1. Introduction

Rice is the staple food for more than half of the world’s population, and stable rice production is central to global food security [1]. Seed vigor, particularly the speed and uniformity of germination, directly determines the efficiency of seedling establishment and the yield potential of the crop. Given the widespread adoption of simplified cultivation methods and the increasing instability of growing seasons due to climate change, breeding varieties with rapid and uniform germination has become a key objective to enhance crop adaptability [2].

Seed germination is a precisely regulated process marking the transition from dormancy to active growth. The classical theory proposes that it is governed by the antagonistic balance between abscisic acid (ABA) and gibberellin (GA) [3]. ABA maintains dormancy through its signaling pathway, whereas GA relieves repression and promotes germination by triggering the degradation of DELLA proteins, such as SLR1 in rice [4]. This hormonal paradigm has provided a foundation for the field. However, current knowledge faces notable limitations. On the one hand, the vast majority of mechanistic studies rely on large-effect mutants in Arabidopsis or rice, failing to systematically uncover the global regulatory basis underlying the continuous natural variation in germination speed across a diverse rice germplasm. On the other hand, research has predominantly focused on transcriptional regulation and a limited set of phytohormones, whereas the dynamic network of metabolites that plays critical roles in this process remains poorly understood [5,6]. As direct executors of biological activities and signaling molecules, metabolites likely constitute a crucial layer for the fine-tuning of germination rate beyond the classical phytohormone framework [7].

Although phytohormones are central regulators, phytohormone signals often perform functions by reshaping metabolic networks, and specific metabolites (such as amino acids) can in turn regulate phytohormone or energy states (such as the TOR pathway) [8,9]. Recent multi-omics studies have provided new perspectives for unraveling the mechanisms underlying seed germination rate [10]. Integrated metabolomic and proteomic analyses can directly capture functional metabolites and their enzymatic basis that determine germination efficiency. These investigations have revealed that rapid germination is closely associated with the coordinated activation of multiple pathways, including amino acid metabolism, energy supply, and antioxidant defense [11]. Nevertheless, research frameworks that systematically identify causal hub metabolites with regulatory functions within natural populations, and provide experimental validation of their regulatory roles, are currently lacking.

In this study, we adopted a phenotype-based reverse genetics strategy. First, we phenotypically classified 56 genetically diverse rice accessions into RG and DG groups based on their germination speed, comprising 28 accessions in each group. Using integrated metabolomic and proteomic analyses, we systematically identified metabolites specifically enriched in the Rapid Germination group. Among these, the amino acid metabolic pathways exhibited the most pronounced changes. We selected glutamine, the most significantly differential metabolite within this pathway, as a representative target for in-depth functional validation. Exogenous application of glutamine significantly promoted the germination process in Delayed Germination varieties, directly confirming its key causal role in regulating seed germination speed. This study not only provides new evidence for understanding the metabolic regulation of seed germination but also offers potential targets for crop genetic improvement.

2. Results

2.1. Non-Targeted Metabolomics Analysis of Rice Germinated Seeds

To comprehensively investigate the metabolic basis underlying differences in rice seed germination rate [12], we selected 56 natural rice populations with extensive genetic diversity from natural germplasm resources and grouped them based on radicle length at 24 h after germination. These populations were grouped based on the time required for seeds to reach a 50% germination rate, and were classified into the Rapid Germination (RG) group and the Delayed Germination (DG) group [13,14]. Untargeted metabolomic profiling was then performed using HPLC–TOF–MS.

To characterize metabolic differences between RG and DG seeds during this critical germination stage [15], metabolites were extracted and analyzed from pooled samples representing each group. The metabolic spectrum signal of total ions showed a highly complex metabolic profile [16], and there was a big difference between the two groups of RG and DG (Figure 1A). Especially in 0.5–1.5 min and 7–13 min, there was a large difference, respectively, corresponding to a large number of water-soluble and fat-soluble metabolites. In addition, principal component analysis (PCA) was performed on the two groups of RG and DG based on non-target metabolic signals [17]. Principal component analysis showed that PC1 and PC2 explained 44.43% and 27.86% of the variation, respectively (Figure 1B). RG and DG samples were successfully separated from PC1 and PC2, indicating distinct metabolite profiles between the two groups.

2.2. Metabolite Identification Based on Non-Targeted Metabolomics Analysis

To further characterize metabolite differences between the two groups of RG and DG, we used Thermo Compounds Discovery 3.3 software to annotate the non-targeted metabolic spectrum data through the public metabolic database [18]. MS/MS spectra were acquired in a targeted MS2 mode for selected precursor ions to obtain fragmentation patterns. The compounds were identified by purchasing standard products, and similar compounds were speculated and identified according to the fracture rules of the compounds [19].

Based on high-resolution mass spectrometry data, we identified the selected differential metabolites by a combination of accurate mass number matching and secondary mass spectrometry (MS/MS) fragment resolution [20]. Several metabolites such as LysoPC (16:1), L-glutamyl-L-glutamic acid, and Quercetin-5-O-hexoside were used as examples.

For the lipid metabolite LysoPC (16:1) (Figure 2A), the Quasi-molecular ion ([M+H]^+^ m/z 494.3247) detected in the positive ion mode is highly consistent with the theoretical value (C_24_H_50_NO_7_P, error < 2 ppm). The identification of this compound mainly depends on the fact that the two characteristic fragment ions m/z 184.0745 (choline phosphate head group) in the MS/MS spectrum are diagnostic ions for lysophosphatidylcholines [21]. And m/z 311.2529 ([M+H-183.0718]^+^) was formed by the loss of 183.0718 Da (choline phosphate fragment), corresponding to [M+H-183.0718]^+^, suggesting that the fatty acid chain is hexadecenoyl (16:1). The mechanism of these fragments is completely consistent with the standard cleavage path of LysoPC (Figure 2B,C), thus confirming that the structure of the compound is LysoPC (16:1). According to similar fragmentation patterns, we annotated 19 LPCs and found that LPC accumulated more in seeds with faster germination.

For the dipeptide metabolite L-glutamyl-L-glutamic acid ([M+H]^+^ m/z 277.1036) and the theoretical value (C_10_H_17_N_2_O_7_, error < 2 ppm), the compound produced a series of fragments in the MS/MS spectrum that conformed to the dipeptide cleavage rule (Figure 2D). Among them, m/z 147.0532 ([Y^0^]^+^) represents the glutamine residue at the C-terminus, and m/z 84.0458 ([Z^1^]^+^) is derived from the neutral loss and rearrangement accompanying the peptide bond cleavage (Figure 2E) [22], the appearance of these fragments and their relative intensity distribution. It is consistent with the cleavage behavior of glutamine reported in the literature, which supports the identification of the compound as L-glutamyl-L-glutamic acid (Figure 2F). In a similar way, we also annotated dipeptides such as γ-glutamine, aspartic acid-L-proline, aspartic acid-lysine, asparagine-phenylalanine, histidine-leucine, glycine-lysine, tyrosine-asparagine, and tyrosine-glutamine, which were enriched in seeds with Delayed Germination.

The flavonoid glycoside compound Quercetin-5-O-hexoside ([M+H]^+^ m/z 465.1033) is highly consistent with the theoretical value (C_21_H_21_O_12_, error < 3 ppm) (Figure 2G). The key identification of the compound is based on the significant glycosidic bond cleavage observed in the MS/MS spectrum [23], that is, the neutral loss of the hexosyl group (162.0588 Da) of the excimer ion to generate a highly abundant aglycone fragment ion m/z 303.0505 ([Y^0^]^+^), which is further confirmed to match the fragment spectrum of the quercetin standard (Figure 2H). This cleavage mode (preferential loss of glycosyl group) is a typical feature of Flavonoid-O-glycoside. Combined with its precise mass, it can be confirmed that the compound is Quercetin-5-O-hexoside (Figure 2I). Based on similar fragmentation patterns, we annotated 30 Flavonoid-O-glycosides, and found that Flavonoid-O-glycosides accumulated more in seeds with slower germination.

In addition to the above representative metabolites, this study adopted a unified standard for all identified metabolites: The first-level mass error was controlled within 5 ppm, and the secondary mass spectrometry fragments were required to match the standard fragmentation spectra, public databases (such as HMDB and MassBank), and the fragmentation rules in the published literature, or to speculate and identify similar compounds according to the fracture rules of the compounds. This process ensures the high confidence of metabolite annotation and lays a reliable data foundation for subsequent biological interpretation.

2.3. Targeted Metabolomics Bioinformatics Analysis

To analyze the accumulation patterns of metabolites in the two groups of RG and DG, we quantitatively detected the content of 1386 metabolites annotated in these 56 varieties by LC-MS. Among the 1386 metabolites are 381 lipids, 169 amino acids and their derivatives, 135 flavonoids, 134 polyamines, 101 organic acids and their derivatives, 71 terpenoids, 53 phytohormones, 44 alkaloids, 36 vitamins and some other compounds (Figure 3A). First, we conducted a correlation analysis of the test results (Supplementary Figure S1). The repeatability of the samples between the three biological replicates was good, and the correlation between the two groups of RG and DG was high. Then the PCA was further carried out. The results showed that the varieties with Rapid Germination gathered together, and the varieties with DG gathered together (Figure 3B). The specificity of germination speed was greater than that of variety specificity [24]. Secondly, we performed hierarchical cluster analysis on the content of these 1386 metabolites identified in 56 different rice varieties. The results showed that the metabolites had different enrichment patterns between the two groups of RG and DG. The content of amino acids and their derivatives, phytohormones and some other substances in the group of Rapid Germination was significantly higher than that in the group of DG (Figure 3C). To visualize the metabolic enrichment pattern between the two groups of RG and DG, we generated a volcanic map by screening differential metabolites. According to the screening threshold, we successfully identified 264 differential metabolites. Among them, 198 metabolites were significantly up-regulated in the RG group (Supplementary Figure S2, log_2_ FC > 0.5 and −log_10_ P > 1.3), and 66 metabolites were significantly up-regulated in the DG group (Supplementary Figure S2, log_2_ FC < −0.5 and −log_10_ P > 1.3). It is worth noting that, among the up-regulated metabolites in the RG group, several points are located in the upper right region of the graph. These points are glutamine, aspartic acid, lysine, reduced glutathione, jasmonic acid-L-isoleucine, jasmonic acid-L-valine, etc. They also have extremely high statistical significance and large expression multiples, which are the key candidate targets for subsequent functional verification.

Detection and identification of specific LysoPC (16:1), L-glutamic acid-L-glutamine, and Quercetin-5-O-hexoside metabolite signals by Q-Exactive LC-MS/MS. (A) EIC (extracted ion chromatogram) of GR0715 at 8.6 min. (B) MS/MS spectra of GR0715 at m/z 494.3247, and the metabolite was identified as LysoPC (16:1). (C) The molecular structure of the LysoPC (16:1) and its general fragmentation rules. (D) EIC (extracted ion chromatogram) of GR0143 at 0.8 min. (E) MS/MS spectra of GR0143 at m/z 277.1036, and the metabolite was identified as L-glutamic acid-L-glutamine. (F) The molecular structure of the L-glutamic acid-L-glutamine and its general fragmentation rules. (G) EIC (extracted ion chromatogram) of GR0419 at 4.1 min. (H) MS/MS spectra of GR0419 at m/z 271.0981, and the metabolite was identified as Quercetin-5-O-hexoside. (I) The molecular structure of the Quercetin-5-O-hexoside and its general fragmentation rules. Std, standard; RG, Rapid Germination; DG, Delayed Germination.

Summary of metabolic profiling and tissue variability analysis. (A) Categorical pie chart of the 1386 annotated metabolites. (B) PCA results of the metabolome data from Rapid Germination (RG), Delayed Germination (DG) and quality control (QC) samples. (C) Hierarchically clustered heatmap of the 1386 annotated metabolites from 56 germinated seed samples.

2.4. Proteomic Analysis

To analyze the regulatory basis of rice seed germination rate difference at the protein level, we performed proteomics analysis on the germination seeds of these 56 samples [25], and a total of 7348 quantifiable proteins were identified. PCA was performed on the protein accumulation levels of all samples. The results showed that the samples of the RG group and the DG group showed a significant separation trend, indicating that the two groups of materials had essential differences in the overall protein accumulation profile (Figure 4A). Individual samples gathered closely in the group, reflecting good biological repeatability. Then we used FC > 2 or FC < 0.5 as the threshold, and a total of 442 significantly differentially accumulated proteins (DAPs) were screened. Among them, 340 proteins were significantly up-regulated in the RG group, and 102 proteins were significantly up-accumulated in the DG group. Simultaneously, the accumulation levels of OsLEA3-1 (LOC_Os05g46480) showed no significant difference between the DG and RG groups. To further reveal the expression patterns of DAPs, we performed a hierarchical cluster analysis based on 340 protein accumulation levels that were significantly up-accumulated in the RG group and 102 protein accumulation levels that were significantly up-accumulated in the DG group. The heat map showed that all samples were accurately divided into two clusters, corresponding to the phenotypic classification of the RG group and the DG group (Figure 4B). At the same time, proteins are also clustered into several modules with co-expression characteristics. For example, the protein modules highly accumulated in the RG group are mainly enriched in energy metabolism, protein biosynthesis and other pathways, and the protein modules highly accumulated in the DG group were related to stress response and dormancy maintenance.

2.5. Enrichment Analysis of Differential Metabolites and Proteins

Based on the joint KEGG pathway enrichment analysis of differential proteins and differential metabolites, we obtained a key metabolic network map closely related to the germination rate of rice seeds (Figure 5). The results showed that amino acid metabolism, energy supply and antioxidant balance were the core biological processes to distinguish the RG and DG groups. The most significantly enriched pathway is alanine, aspartic acid and glutamic acid metabolism, which directly points to the key metabolite of functional verification in this study, glutamine, and confirms its pivotal position in germination regulation from the system level. At the same time, the glutathione metabolic pathway was also significantly enriched, suggesting that RG seeds may have stronger reactive oxygen species scavenging ability to maintain redox homeostasis. In addition, the enrichment of related pathways such as starch and sucrose metabolism [26], pyruvate metabolism [27], and TCA cycle [28] jointly indicated that RG seeds initiated a more efficient carbon skeleton supply and energy generation program in the early stage of germination. These data reveal from the perspective of integrative omics that rapid seed germination is driven by a synergistic metabolic network centered on amino acid metabolism such as glutamine, which simultaneously strengthens energy production and cellular antioxidant protection, thus providing sufficient material and energy and a stable cell environment for the rapid elongation of radicle and germ.

2.6. Exogenous Glutamine Treatment Promotes the Early Radicle and Plumule Elongation in Delayed Germination Rice Cultivars

To verify the potential function of the screened differential metabolite glutamine in regulating seed germination, we selected three representative varieties (GR178, GR253, and GR467) in the DG group. For each accession, at least 100 seeds were used with three biological replicates. Glutamine solution (treatment group) and an equal amount of sterile water (control group) were used for germination [29,30]. The germination phenotype was observed immediately after 24 h of germination, and the lengths of the radicle and plumule were measured and recorded, respectively. Compared with the control group, exogenous application of glutamine significantly promoted the radicle and plumule elongation of all three test varieties at the early stage of germination (24 h) (Figure 6A). The results showed that glutamine had a positive regulatory role in promoting rice seed germination, especially in initiating radicle growth. Exogenous glutamine treatment significantly promoted the early growth of the three DG varieties, but the response patterns of different varieties were different. Specifically, the radicle and plumule lengths of GR178 increased by 1.9-fold and 1.7-fold, respectively, compared with the control, showing a balanced promotion (Figure 6B,C); the plumule growth of GR253 and GR467 was more strongly stimulated, which increased to 2.7-fold and 2.3-fold of the control, respectively, and the radicle growth was relatively small (1.4-fold and 1.1-fold, respectively) (Figure 6B,C). Collectively, these results indicate that glutamine responsiveness is genotype-dependent and, in two of the three accessions tested, plumule elongation responded more strongly than radicle elongation, suggesting that glutamine may preferentially modulate early plumule development and that varietal differences in downstream physiological or molecular pathways could underlie the observed organ-specific growth responses.

3. Discussion

The seed germination rate is a crucial agronomic trait that determines the survival of rice seedlings, but the underlying metabolic regulatory network still requires further exploration [31]. In this study, by comparing the metabolome and proteome of Rapid Germination and Delayed Germination rice varieties, it was found that amino acid metabolism was significantly enriched in Rapid Germination varieties. Subsequent exogenous functional experiments confirmed that glutamine could selectively and significantly promote the germination of some Delayed Germination varieties, thus establishing its role as a key functional metabolite that positively regulates the germination rate of rice seeds.

We speculate that the germination-promoting effect of glutamine may be achieved through multiple mechanisms. First, as a key precursor of the main nitrogen transport form and protein/purine synthesis, its exogenous supplementation may directly relieve the ‘nitrogen limitation’ of rapid anabolism in Delayed Germination seeds [32]. Secondly, recent studies have shown that glutamine can integrate nutrition and phytohormone signals by affecting TOR kinase activity [33]. Interestingly, the proteomic data of this study preliminarily showed that the abundance of enzyme proteins related to nitrogen assimilation and energy metabolism changed significantly in the varieties responding to glutamine, which provided support for the above hypothesis.

This study combined metabolomics and proteomics to reveal a key metabolite affecting seed germination and also provided new ideas for crop cultivation practice. The findings suggest that glutamine treatment could effectively initiate the germination process and improve the uniformity of field emergence for ‘germination-sensitive’ varieties such as GR253 and GR467. In the future, ‘glutamine responsiveness’ can be used as a physiological index to assist traditional breeding to screen excellent strains [34]. From a broader perspective, this work directly relates metabolite functional verification to crop phenotypic improvement, providing a specific case for metabolic engineering breeding [35,36].

It should be noted, however, that the study primarily focused on early-stage phenotypes under 24 h treatment, and thus does not fully capture the complete dynamics of germination, and there are some limitations in testing only a single glutamine concentration. Future studies should set concentration and time gradients to accurately quantify their effects. More importantly, it is urgent to reveal its mechanism of action at the molecular level: the expression changes in genes related to glutamine metabolism, nitrogen assimilation and cell elongation before and after treatment can be compared by transcriptome analysis [37]; genetic methods can also be used to verify the function of key genes [38]. Ultimately, conducting verification tests in a controlled environment or in the field will be a key step in assessing its practical application value.

4. Conclusions

In this study, we integrated metabolomics, proteomics, and exogenous functional assays to systematically dissect the metabolic regulatory basis underlying natural variation in rice seed germination rate and identified glutamine as a key positive regulator.

First, we constructed a germination rate variation population covering 56 materials and found that the RG seeds were significantly enriched in amino acid metabolism, energy metabolism and antioxidant pathways at the metabolic level. Among them, the content of glutamine was significantly increased in the Rapid Germination group. Subsequent exogenous treatment experiments directly confirmed that glutamine could effectively improve the germination phenotype of some Delayed Germination varieties (such as GR178, GR253 and GR467) and significantly promote the early elongation of radicle and plumule, thus verifying that it was a key functional metabolite that positively regulated the germination rate. It is worth noting that this effect has variety specificity, revealing the genetic complexity of the germination regulatory network.

Second, our integrative results showed that rapid seed germination was not driven by a single metabolite but rather a synergistic metabolic process. The program focuses on amino acid metabolism such as glutamine, and simultaneously activates multiple pathways such as carbon supply (such as starch and sucrose metabolism), energy generation (such as pyruvate metabolism and TCA cycle), and cell protection (such as glutathione metabolism), which provides a complete material, energy and steady-state basis for the rapid growth of embryos.

In summary, this study not only provides new insights into the metabolic regulation mechanism of seed germination but also provides potential targets for crop genetic improvement. Future work can focus on the downstream signaling mechanism of glutamine and the causes of variety specificity [39], and explore its application potential as an environmentally friendly seed treatment agent or molecular marker in breeding.

5. Materials and Methods

5.1. Plant Materials and Sample Collection

A diverse worldwide collection of 56 O. sativa accessions including both landraces and elite varieties was obtained [40]. Information about the accessions, such as variety name, country of origin, longitude and latitude origin and subpopulation identity are listed in Supplementary Table S1. Rice seeds were produced in 2023 at the Huazhong Agricultural University Rice Cultivation Base in Lingshui City, Hainan Province, China. After reaching maturity under natural growth conditions, seeds were harvested. Fifty-six rice varieties were selected, with thirty seeds per variety and three biological replicates per variety subjected to standard soaking and germination induction protocols [41]. Specifically, seeds were organized in perforated soaking bags and placed in an oven at 37 °C for two days for drying [42]. The dried seeds were then transferred to beakers filled with water and maintained in the same oven at 37 °C for 24 h, with the water replaced every 12 h during soaking. Upon observation of plumule emergence, germination was induced by transferring the seeds to Petri dishes lined with moistened cotton and continuing incubation at 37 °C for another 24 h. Immediately following this step, the seeds were placed into centrifuge tubes and divided into two parts. One part was rapidly frozen in liquid nitrogen, and stored at −80 °C, after which they were subjected to freeze-drying. The other part was used for phenotypic examination [43]. To ensure data accuracy, the radicle and plumule lengths of each individual seed were measured and recorded. After the seeds had exposed their white radicles, they were allowed to germinate for another 24 h. Then, using an electronic digital caliper, the lengths were measured. When the lengths of the radicles and plumules were less than 5 mm, they were defined as the Delayed Germination (DG) group; when the lengths were greater than 5 mm, they were defined as the Rapid Germination (RG) group.

5.2. Chemicals and Reagents

Chromatography-grade methanol and acetonitrile used for metabolite extraction from germinating rice seeds and subsequent analysis via ultra-performance liquid chromatography–tandem mass spectrometry (UPLC-MS/MS), as well as the deionized water produced by a laboratory water purification system, were all supplied by Thermo Fisher Scientific (Rockford, IL, USA; http://www.thermofisher.cn/cn/zh/home.html, accessed on 12 January 2025). The internal standard lidocaine used for metabolic sample analysis was purchased from ANPEL Laboratory Technologies (Shanghai) Inc., Shanghai, China (http://www.anpel.com.cn/).

The reagents required for protein extraction from germinating rice seeds, including Tris-HCl, urea, dithiothreitol (DTT), iodoacetamide (IAM), ammonium bicarbonate solution, and trypsin working solution, were all purchased from Sigma-Aldrich, St. Louis, MO, USA (https://www.sigmaaldrich.cn, accessed on 12 January 2025).

Amino acid standards, along with authentic chemical standards for the metabolites LysoPC (16:1), L-glutamyl-L-glutamate, and quercetin-5-O-hexoside, were obtained from Macklin Biochemical Co., Ltd., Shanghai, China (http://www.macklin.cn/, accessed on 12 January 2025).

5.3. Sample Extraction

Samples were ground using a tungsten carbide ball mill (BM 500, Anton Paar, Graz, Austria) at a frequency of 30 Hz for 60 s. Subsequently, 0.05 g of the ground sample was weighed using an electronic balance and transferred into a 2 mL centrifuge tube.

For metabolite extraction [44], a 70% methanol solution (containing 10 ppb lidocaine as an internal standard) was added to the samples at a solvent-to-sample ratio of 8:1 (v/w). The sample was vortexed (Scientific Industries) for 10 s and then incubated at 4 °C for 10 min, and this process was repeated three times. Subsequently, the samples were maintained at 4 °C overnight to ensure thorough extraction. The next day, the sample was centrifuged at 12,000 rpm for 10 min at 4 °C, and the supernatant was collected and filtered into brown injection vials equipped with inner cannulas. Samples were vortexed for 10 s immediately prior to LC-MS analysis or stored at −80 °C for long-term stability.

The process of protein extraction and enzymatic hydrolysis is briefly described as follows: After grinding and crushing the plant samples, the protein was extracted using 8 M urea extract. After ultrasonic crushing and centrifugation, the supernatant was obtained, and the protein concentration was determined by the Bradford method. For each sample, 50 ug protein solution was placed in an ultrafiltration tube (30 KD, Millipore), followed by reduction and alkylation with dithiothreitol (DTT) and iodoacetamide. After reductive alkylation, the solution was washed twice with 8 M urea and replaced with 100 mM ammonium bicarbonate for subsequent enzymatic hydrolysis. Trypsin was added at a ratio of 1: 50 and digested overnight at 37 °C. Finally, the peptide solution was collected by centrifugation. After vacuum drying, the peptides were dissolved in 0.1% trifluoroacetic acid (TFA) buffer and desalted by C18, and stored at −20 °C or for subsequent analysis.

5.4. Establishment of a Targeted and Non-Targeted Metabolomics Method

The non-targeted metabolomics data were matched by Thermo Compounds Discovery 3.3 software, and the calculated metabolite molecular formula was confirmed by purchasing the corresponding standard after the Chemspider database (http://www.chemspider.com/, accessed on 12 January 2025) query comparison. For metabolites without standard substance or molecular formula, the structure was speculated according to the secondary spectrum of metabolites, and identified by comparing with the standard substance m/z, secondary spectrum and retention time with similar structure. For metabolites that cannot be identified by the above two methods, they were identified by the publicly available literature and publicly available mass spectrometry databases, including HMDB, MassBank, METLIN, etc. [45].

Based on UPLC-MS/MS, the metabolic database of rice germination seeds was constructed, and the detection results of metabolic samples were analyzed by extensive targeted metabolic detection [45,46]. HPLC-ESI-Q-Eactiv-MS/MS (Thermo Sicentific Q-Eactiv, Thermo Fisher Sicentific, Rockford, IL, USA) was used for non-targeted analysis. LC-ESI-QTRAP-MS/MS) AB Sciex 6500, Applied Biosystems, Carlsbad, CA, USA) multiple reaction monitoring (MRM) mode was used for targeted analysis of metabolites. Analyst 1.6 software (AB Sciex) was used for data acquisition and peak integration calculation. The compound quantification method is to calculate the area of each peak and compare it with the standard curve of the lidocaine standard.

The conditions of liquid chromatography were as follows: the mobile phase, water phase was ultrapure water (adding 0.04% acetic acid), organic phase was acetonitrile (adding 0.04% acetic acid); elution gradient: water: acetonitrile, 0 min: 95: 5 v/v, 12.0 min: 0: 100 v/v, 13.2 min: 0: 100 v/v, 13.3 min: 100: 0 v/v, 15.0 min: 95: 5 v/v; the flow rate was 0.40 mL/min. The column temperature was 40 °C. The injection volume was 2 μL. The chromatographic column was Shim-pack GISS C18 (pore size 1.9 μm, dimensions 2.1 × 100 mm).

5.5. DIA Protein Detection Method

High performance liquid chromatography (HPLC) analysis was performed using Vanquish Neo UHPLC (Thermo Fisher Scientific, Rockford, IL, USA) [47]. Liquid phase program: 16.5 min, flow rate: 500 nL/min. Phase A: 0.1% FA aqueous solution, phase B: 80% ACN aqueous solution containing 0.1% FA. The effective separation gradient of the liquid phase was set as 0–2 min, 1–11% B; 2–9.5 min, 11–23% B; 9.5–14 min, 23–35% B; 14–16.5 min, 99% B.

Proteomics analysis was performed using an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific, Rockford, IL, USA) [48]. Ion Opticks’s fourth-generation Aurora Series^®^ ELITE was used as the chromatographic column (C18 analytical column, IonOpticks, Melbourne, Australia). The C18 conventional quantitative proteomics column specification was 15 cm × 75 μm. The acquisition mode was positive ion mode, the spray voltage was 2.2 kV, and the ion transport tube temperature was 320 °C. The data acquisition mode adopts the independent data acquisition mode (DIA) [49]. Mass-to-charge ratio scan range: 380–980, resolution: 240,000, standardized automatic gain control (AGC) parameters: 300%, maximum ion implantation time: 3.5 ms. Secondary mass-to-charge ratio scanning range: 150–2000, resolution: 120,000, standardized automatic gain control (AGC) parameters: 500%, the maximum ion implantation time is set to Auto. Isolation window: 2 Da, collision energy parameter: 29%. Each sample collected 500 ng of peptides per needle for mass spectrometry analysis, three biological replicates per sample, and one QC mix per 10 samples as a reference for the stability of the entire system.

5.6. Sample Repeatability Analysis

The ‘corrplot’ package in R 4.5.0 (http://www.cran.r.project.org/, accessed on 12 May 2025) was utilized to analyze the sample repeatability [50], and the ‘pheatmap’ package in R 4.5.0 was used to draw the correlation heatmap [51].

5.7. Screening of Differential Metabolites and Proteins

When the content of a metabolite and protein meets the p-value of Student’s T test ≤ 0.05 and the difference between the metabolite or protein in the RG group and the DG group is Fold Change ≥ 2, it is considered that this metabolite or protein has a significant difference. The selected differential metabolites and proteins were plotted into volcanic maps using OmicStudio (https://www.omicstudio.cn/, accessed on 12 May 2025) to visually display the results.

5.8. Differential Metabolism and Protein Enrichment Analysis

Enrichment analysis was performed on the differential metabolism and proteins of the two groups of samples with Rapid Germination and Delayed Germination. In the enriched metabolic pathways, when p < 0.05 was considered significant, the bubble diagram was drawn using the R 4.5.0 package ‘ggplot2’ to visually display the KEGG enrichment results.

5.9. Functional Verification of Differential Metabolites

The full seeds of three representative varieties (GR178, GR253, and GR467) were selected from the Delayed Germination varieties screened in the previous period for verification. The glutamine treatment group and the sterile water control group were set up in the experiment. Each variety in each group contained at least 50 seeds and 3 biological replicates. The specific steps were as follows: firstly, the Petri dish filter paper with disinfected seeds was infiltrated with 5 mM glutamine solution (treatment group) or the same amount of sterile water (control group). Subsequently, the culture dish was placed in dark conditions at 37 °C for 24 h. Finally, the germinated seeds were randomly selected and immediately frozen in liquid nitrogen for subsequent analysis. The length of the radicle and germ of each seed was measured using an electronic digital caliper, and the germination rate was counted. Data processing used an independent sample T test to compare the treatment differences within the same variety, and used two-way analysis of variance to compare the interaction effects between varieties and treatments. The significance level was set to p < 0.05.